consider refactor to scikit-style mixins

[knaaptime]

@renanxcortes and I have had some conversations about refactoring this project so that, instead of each segregation index being implemented as its own class, we might have two classes (e.g. spatial and aspatial), and the indices themselves would move to subclasses or functions.

In another thread, as we work on spopt we're making a concerted effort to use the BaseClass/Mixin structure that scikit-learn uses. I want to raise the option here of adopting a similar architecture, which i think would make a lot of sense for this project

[renanxcortes]

Yes, I totally agree that refactoring like that would be much better! This is something that I'll start shortly in a branch!

[renanxcortes]

@knaaptime I recently update this project to this new style! I'm still working on some other improvements, but I think we can close this issue, right?

[knaaptime]

Yup, nice work

[renanxcortes]

I'm reopening this issue, since the current version is an improvement of the subclasses with the type:

index_type.index_name

But not as the Mixins type mentioned in the first comment that scikit-learn does. Such as:

https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/cluster/affinity_propagation_.py

[knaaptime]

I'm eyeballing this refactor sometime in the near future. What i'd like to do is target this as segregation 2.0 because the api will change pretty substantially. What I have planned is to do away with the spatial/aspatial distinction in the module and make it clear that [almost] any of the indices can be spatialized by passing a W or a Network to the init.

A more meaningful distinction will come in the form of two base classes: SingleGroupIndex and MultiGroupIndex, each of which can be spatial or aspatial. Currently this is a muddy division in the library, with "aspatial" indices divided into single and multigroup modules and spatial indices all lumped together.

the new modules will be

indices:
singlegroup
multigroup
local

analysis
decomposition
inference

Utility (subsumes network and compute_all)
util

[knaaptime]

from segregation.singlegroup import GiniSeg
from segregation.multigroup import MultiGiniSeg

g_a = GiniSeg(group_pop_var, total_pop_var)   # aspatial single group
g_s = GiniSeg(group_pop_var, total_pop_var, W)  # spatial single group

mg_a = MultiGiniSeg(input_df, groups_list)  # aspatial multigroup
mg_s = MultiGiniSeg(input_df, groups_list, Network)  # spatial multigroup

[knaaptime]

this will make it much easier to accomodate #104 and will also make the decomposition easier to handle for spatial indices